Expert ensemble musicians produce exquisitely coordinated sounds, but rehearsal is typically required to do so. Ensemble coordination may thus be influenced by the degree to which individuals are familiar with each other's parts. Such familiarity may affect the ability to predict and synchronize with co-performers' actions. Internal models related to action simulation and anticipatory musical imagery may be affected by knowledge of (1) the musical structure of a co-performer's part (e.g., in terms of its rhythm and phrase structure) and/or (2) the co-performer's idiosyncratic playing style (e.g., expressive micro-timing variations). The current study investigated the effects of familiarity on interpersonal coordination in piano duos. Skilled pianists were required to play several duets with different partners. One condition included duets for which co-performers had previously practiced both parts, while another condition included duets for which each performer had practiced only their own part. Each piece was recorded six times without joint rehearsal or visual contact to examine the effects of increasing familiarity. Interpersonal coordination was quantified by measuring asynchronies between pianists' keystroke timing and the correlation of their body (head and torso) movements, which were recorded with a motion capture system. The results suggest that familiarity with a co-performer's part, in the absence of familiarity with their playing style, engenders predictions about micro-timing variations that are based instead upon one's own playing style, leading to a mismatch between predictions and actual events at short timescales. Predictions at longer timescales—that is, those related to musical measures and phrases, and reflected in head movements and body sway—are, however, facilitated by familiarity with the structure of a co-performer's part. These findings point to a dissociation between interpersonal coordination at the level of keystrokes and body movements.

In our daily lives, auditory stream segregation allows us to differentiate concurrent sound sources and to make sense of the scene we are experiencing. However, a combination of segregation and the concurrent integration of auditory streams is necessary in order to analyze the relationship between streams and thus perceive a coherent auditory scene. The present functional magnetic resonance imaging study investigates the relative role and neural underpinnings of these listening strategies in multi-part musical stimuli. We compare a real human performance of a piano duet and a synthetic stimulus of the same duet in a prioritized integrative attention paradigm that required the simultaneous segregation and integration of auditory streams. In so doing, we manipulate the degree to which the attended part of the duet led either structurally (attend melody vs. attend accompaniment) or temporally (asynchronies vs. no asynchronies between parts), and thus the relative contributions of integration and segregation used to make an assessment of the leader-follower relationship. We show that perceptually the relationship between parts is biased towards the conventional structural hierarchy in western music in which the melody generally dominates (leads) the accompaniment. Moreover, the assessment varies as a function of both cognitive load, as shown through difficulty ratings and the interaction of the temporal and the structural relationship factors. Neurally, we see that the temporal relationship between parts, as one important cue for stream segregation, revealed distinct neural activity in the planum temporale. By contrast, integration used when listening to both the temporally separated performance stimulus and the temporally fused synthetic stimulus resulted in activation of the intraparietal sulcus. These results support the hypothesis that the planum temporale and IPS are key structures underlying the mechanisms of segregation and integration of auditory streams, respectively.